![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Metabolism and Pharmacokinetics (H.Z., W.-C.S.), Biotransformation (D.Z., W.L., M.Y., W.G.H.), Discovery Analytical Sciences (C.D., Y.-X.L.), and Discovery Chemistry (W.R.E., Z.G., Y.Z., N.M.), Pharmaceutical Research Institute, Bristol-Myers Squibb Company, Princeton, New Jersey
(Received September 6, 2006; accepted February 12, 2007)
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
). In humans, P450 enzymes are responsible for oxidative metabolism of not only endogenous substrates such as steroids, fatty acids, and bile acids, but also xenobiotics including a majority of drugs, food components, and environmental pollutants (Guengerich, 1995; Wrighton et al., 1996). Among the many human P450s, the CYP3A4 enzyme is by far the most important drug-metabolizing enzyme and accounts for the metabolism of more than 50% of marketed drugs (Guengerich, 1995; Wrighton et al., 1996). In a typical human liver, CYP3A enzymes may represent up to 28% of the total P450 proteins (Shimada et al., 1994; Rendic and Di Carlo, 1997). To date, four members of human CYP3A enzymes (CYP3A4, CYP3A5, CYP3A7, and CYP3A43) have been described, and CYP3A4 is the most abundant hepatic and intestinal form among them. CYP3A5 may account for up to 50% of the total CYP3A proteins in certain individuals and thus potentially plays a role in CYP3A-mediated metabolism and clearance of certain drugs in such populations (Kuehl et al., 2001). CYP3A7 expression is mainly confined to fetal livers (Schuetz et al., 1994; Lamba et al., 2002), and CYP3A43 has been shown to be present at only 0.2 to 5% of the hepatic CYP3A4 content (Gellner et al., 2001).
CYP3A4 enzyme activity is widely variable in the human population because of a combination of genetic and environmental factors, and the variability in CYP3A4 activity can be further exacerbated in the presence of enzyme inhibitors or inducers. This property, along with the major role the enzyme plays in the clearance of many drugs, makes metabolism by CYP3A4 a key issue in drug discovery and development. P450 substrates, particularly those metabolized solely by CYP3A4 with high intrinsic clearance, often exhibit low systemic exposure and variable pharmacokinetics. Clinical development of sole CYP3A4 substrates may require extensive investigations with respect to drug-drug interactions (Gibbs and Hosea, 2003). Exposure of a CYP3A4 substrate can be affected by altered CYP3A4 metabolism, either via a decrease in metabolic clearance (e.g., enzyme inhibition) or an increase in metabolic efficiency (e.g., enzyme induction) (Guengerich, 1997b). It has been shown that compounds depending on CYP3A4 metabolism for the majority of their clearance have a far greater potential for drug-drug interactions than compounds that are metabolized by multiple enzymes when coadministered with CYP3A4 inhibitors (Wrighton et al., 2000; Williams and Feely, 2002). For example, terfenadine, a sole CYP3A4 substrate, displays less than 5% bioavailability in humans and suffers profound drug-drug interaction with ketoconazole (Honig et al., 1993). The increase in systemic exposure of terfenadine in the presence of ketoconazole was found to be associated with dangerous QTc prolongation (Kivisto et al., 1994), and as a result the drug was withdrawn from the market.
Because of the prevalence of drug-drug interactions associated with CYP3A4-mediated metabolism, it has been suggested that human CYP3A4 substrates, especially those solely metabolized by the enzyme, should be avoided as drug candidates (Boxenbaum, 1999). Although such a task is challenging and sometimes practically impossible, concerted efforts have been made in the pharmaceutical industry to avoid sole CYP3A4 substrates during drug candidate optimization. Ideally, compounds with superior drug-like properties, such as having multiple enzymes involved in their metabolic clearance, could be identified or optimized via in vitro and in vivo evaluations.
|
, 2005). During the candidate optimization phase in search for dual-acting receptor antagonists, CYP3A4 inhibition was initially identified as a liability for a number of compounds. In addition, many early structural analogs suffered from high metabolic clearance in liver microsomes. BMS-1 (Fig. 1) was selected as a lead template because of its weak inhibition of P450 enzymes. However, in vitro metabolism studies revealed that the compound was subject to high metabolic clearance in monkey and human liver microsomes. As well, CYP3A was the major P450 enzyme responsible for its in vitro metabolism. BMS-1 also showed highly variable exposure in cynomolgus monkeys after p.o. administration. The present report summarizes an integrated strategy using in vitro metabolism studies and in vivo pharmacokinetic evaluation during the drug discovery stage to modulate CYP3A-mediated metabolism and improve the overall metabolic profile of new analogs. |
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Microsomal Incubations. Incubations with monkey and human liver microsomes were conducted with 1 mg/ml microsomal proteins and 10 µM test compounds at 37°C in the presence of 1 mM NADPH. The reaction was initiated with the addition of NADPH. Aliquots (200 µl) of reaction mixtures were taken at 0, 5, and 30 min time points, and reactions were quenched with the addition of an equal volume of acetonitrile. After centrifugation to remove proteins, supernatants (100 µl) were analyzed by HPLC/UV (
= 280 nm) for metabolic stability determination and by liquid chromatography/tandem mass spectrometry (LC/MS/MS) for metabolite profiling.
Metabolic stability of the test compounds in human and monkey liver microsomes was compared based on the loss (or substrate depletion) of the parent compounds before and after structural modification (halogenation). The substrate depletion approach has been described in the literature for the estimation of kinetic parameters (Obach and Reed-Hagen, 2002; Jones and Houston, 2004; Mohutsky et al., 2006; Nath and Atkins, 2006). The approach is often used during the discovery stage when synthetic standards of metabolites are not available. Because the primary purpose of in vitro microsomal studies was to evaluate whether structural modification would alter metabolic stability, a full kinetic evaluation using the substrate depletion approach was not conducted for the present report. A substrate concentration of 10 µM was used in the present study to ensure 1) adequate detection by LC/UV, and 2) generation of sufficient metabolites for structural identification by LC/MS/MS. It is recognized that Km values of CYP3A-mediated metabolism vary widely, and the 10 µM value may be above those enzymatic reactions; however, the 10 µM concentration used would be still valid for metabolic stability comparison. The approach is also supported by the fact that the efficacious plasma Cmax of these compounds would likely be in the double-digit micromolar range (Murugesan et al., 2005). The metabolite formation after 30 min of incubation was expressed as metabolite/parent ratio based on relative percentile of a specific metabolite in total drug-related UV peaks at 280 nm. Although LC/UV detection may not be as specific as LC/MS/MS, the method has been found to be adequate in semiquantitative measurement of metabolites based on the assumption that structural modifications of the isoxazole ring do not significantly alter the molar extinction coefficient of a compound and its metabolites bearing the biphenyl system (Fura et al., 2003).
Chemical inhibition studies were conducted in liver microsomes under the same conditions as described above, except with the addition of ketoconazole (final concentration of 2 µM). The selectivity of ketoconazole against CYP3A activity is dependent on various factors, such as protein concentrations, probe substrates, liver microsomal preparations, and species (Eagling et al., 1998; Zhang et al., 2002; Carr et al., 2006). Because high concentrations of ketoconazole (>10 µM) inhibit multiple human P450 enzymes (Eagling et al., 1998; Zhang et al., 2002), a concentration of 2 µM was used in the present study. Typically, incubation mixtures without NADPH were preincubated for 5 min in a water bath at 37°C. The reaction was initiated with the addition of NADPH and carried out for an additional 5 or 10 min. The reaction was quenched with 1 volume of acetonitrile. After centrifugation, samples were subject to HPLC/UV (or LC/MS) analysis to assess the inhibition of CYP3A-mediated metabolism.
Incubations with cDNA-Expressed Enzymes. Incubations with each individual P450 enzymes were conducted at 37°C in the presence of 20 pmol/ml P450 proteins, 1 mM NADPH, and 10 µM test compound. Incubations were initiated with the addition of NADPH and carried out for 30 min. Reactions were quenched with the addition of an equal volume of acetonitrile. After centrifugation, supernatants (100 µl) were analyzed by HPCL/UV and LC/MS/MS for metabolite profiling.
Metabolite Profiling and Identification. The incubation mixtures of BMS compounds with liver microsomes or cDNA-expressed P450 enzymes after protein precipitation with acetonitrile were profiled with HPLC/UV and LC/MS/MS. Chromatographic separation was carried out with a Shimadzu Class VP HPLC system equipped with two pumps (model LC-10AT), an autoinjector (SIL 10AD), and a diode array detector (SPD-M10A). A Zorbax RX C-18 column (2.1 x 150 mm, 5 µm) was used along with a C-18 guard column. HPLC was performed with the column enclosed in an Eppendorf CH-30 column heater (Eppendorf North America, Westbury, NY) maintained at 35°C. A linear gradient was employed with two solvent systems, A and B, and a flow rate of 0.4 ml/min. Solvent A consisted of 5% acetonitrile/95% water/0.1% trifluoroacetic acid (v/v), and solvent B consisted of 95% acetonitrile/5% water/0.1% trifluoroacetic acid (v/v). HPLC analysis was initiated with 100% solvent A and 0% B, and held for 3 min followed by a linear gradient over 25 min to a final solvent composition of 10% solvent A and 90% solvent B. Parent compounds and their metabolites were monitored at 280 nm. Peak purity was regularly examined, and if necessary a baseline subtraction was performed using samples at 0 min. Baseline separation was achieved for parent compounds and their metabolites.
Mass spectral analysis was performed with a Finnigan LCQ-deca mass spectrometer (Finnigan MAT, San Jose, CA). An HPLC system (Shimadzu Class VP) was interfaced to the mass spectrometer through an electrospray ionization probe, and sample analysis was conducted in the positive ionization mode. LC eluent was directed to the mass spectrometer via a divert valve with the eluent from 0 to 5 min diverted into waste. The inlet capillary of the mass spectrometer was maintained at 23°C, and nitrogen gas flow, spray current, and voltages were optimized for maximal sensitivity. MS/MS analysis was performed at 20 to 40% of the default collision energy.
Synthesis and Structural Determination of the Major Metabolite of BMS-1. To further define the site of hydroxylation on the dimethyl isoxazole ring of BMS-1, the major metabolite (M3) of BMS-1 was synthesized using a large-scale incubation with monkey liver microsomes. Briefly, a 40-ml mixture containing 0.1 M potassium phosphate buffer (pH 7.4), 10 mM MgCl2, 0.2 mM BMS-1, 4 mM NADPH, and 2 mg/ml protein of pooled cynomolgus monkey liver microsomes in a 100-ml flask was incubated at 37°C for 2 h. The incubation mixture was extracted with ethyl acetate (2 x 40 ml). The ethyl acetate extract was evaporated to dryness with a rotavapor, and the residue was dissolved in 1 ml of dimethyl sulfoxide (DMSO). After removal of insoluble material with centrifugation (13,000 rpm, 3 min), the DMSO solution was subjected to semiprep HPLC separation with the following conditions: 1) column, YMC Pro C-18, 20 x 150 mm, S5 (Waters, Milford, MA); 2) solvents for mobile phases, 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B); 3) gradient, 20 to 38% B in 4 min, 38 to 45% B in 16 min, and 45% for 20 min; and 4) flow rate, 10 ml/min. Fractions containing M3 were pooled and lyophilized. Approximately 3 mg of M3 was obtained as a white solid.
NMR analysis of M3 was performed using a Varian INOVA 500 MHz NMR spectrometer (Varian, Walnut Creek, CA). A solution of M3 (1.5 mg in 220 µl of DMSO-d6) was transferred to a 3-mm NMR tube, and 1H, 13C, 1H-1H-gradient-selected correlation spectroscopy, 1H-13C-heteronuclear multiple quantum coherence, 1H-13C-heteronuclear multiple bond coherence (HMBC), and rotating frame nuclear Overhauser enhancement spectroscopy NMR spectra were acquired at room temperature. For reference, NMR spectra of BMS-1 (formic acid salt in DMSO-d6) were also acquired.
Monkey Pharmacokinetics. Male cynomolgus monkeys (body weight 4–5 kg) were obtained from Charles River Labs (Houston, TX) and housed individually. Vascular access ports were surgically implanted into the femoral vein and femoral artery. All the experimental procedures were approved by the BMS Institutional Animal Care and Use Committee. Monkeys were fasted overnight, and water was provided ad libitum throughout the study. Food (Harlan Teklad monkey chow, Madison, WI) was allowed at approximately 4 h postdose. The dosing vehicle consisted of 10% ethanol/40% polyethylene glycol 400/50% distilled water (v/v). A minimum of 10% ethanol was needed to provide a clear dosing solution of test compounds. Monkeys received test compounds at 10 µmol/kg dose level via oral gavage, and the final dosing volume was 1 ml/kg. The amount of ethanol given did not produce any abnormal behaviors in these monkeys. EDTA was used as the anticoagulant, and an approximately 0.5-ml aliquot of blood was collected at each time point (15, 30, and 45 min, 1, 2, 4, 6, 8, and 10 h postdose). Plasma samples were obtained by centrifugation at 4°C and 3000 rpm for 15 min.
To assess in vivo CYP3A activity, midazolam was given p.o. at 1.5 mg/kg to the same three monkeys, and plasma samples were obtained in a similar fashion as described above.
Plasma drug concentrations were determined by validated LC/MS/MS methods. Each individual method was optimized for a given analyte, and standard curves typically covered a linear range of 1 to 5000 ng/ml. Quality control samples within the concentration range were prepared and analyzed, and values for the quality control samples were normally within ± 15% of nominal concentrations. Internal standard (IS) was alternated depending on the analyte, e.g., BMS-1 (IS) for compounds BMS-2 and BMS-3 (analytes), and BMS-2 (IS) for BMS-1 (analyte).
Pharmacokinetic parameters were estimated based on plasma concentration versus time data by noncompartmental methods using the KINETICA software program (version 2.4, InnaPhase Corporation, Philadelphia, PA). The Cmax and Tmax values were recorded directly from experimental observations. Values of area under the curve (AUC0-t) were estimated using a combination of linear and log trapezoidal summations.
|
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
Incubations of BMS-1 with cDNA-expressed individual human P450 enzymes were conducted to determine which P450 enzyme was responsible for its metabolism. Among the five major human P450 enzymes (CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4) tested, CYP3A4 was the primary enzyme that metabolized BMS-1 (Table 2). Turnover of BMS-1 by other P450 enzymes was too low to detect any appreciable metabolism. HPLC/UV analysis showed that M3 was formed in the CYP3A4 incubation, along with several other minor metabolites (data not shown). In human and monkey liver microsomes, the metabolism of BMS-1 was inhibited by greater than 95% in the presence of ketoconazole (a CYP3A4 inhibitor). Although there are no published detailed studies of the relationship of ketoconazole concentration to inhibition of individual cynomolgus P450 enzymes, the assumption was made based on similarities to human P450s that the inhibition at 2 µM in cynomolgus liver microsomes would be predominately of CYP3A enzymes. Therefore, BMS-1 was likely a CYP3A substrate for both human and monkey enzymes, leading to the formation of M3.
|
Identification of Metabolic Soft Spots for BMS-1. Mass spectral analyses (MS/MS and/or MS3) were performed on M3 and other metabolites to identify metabolic soft spots of BMS-1. As shown previously (Fig. 3), there were likely three metabolites (M1, M3, and M4) that could be formed via hydroxylation (e.g., M + 16 + 1 peaks). As shown in Fig. 4A, M3 had a fragmentation pattern of m/z 585 
526 428 357 231. The characteristic fragments of m/z of 357 and 231 indicated that M3 could be formed via the hydroxylation of the methyl group on the isoxazole ring. In contrast, M4 followed a fragmentation pattern of m/z 585 526 508 456 341 (Fig. 4B), suggesting that M4 was likely a hydroxylated product of the isopropyl side chain. The mass spectrum of M1 was not conclusive because of background interference.
|
1.64 and 1.68 ppm) had correlations to C3, C4, and C5 on the isoxazole ring, whereas the methyl protons at C10 position (2.17 and 2.19 ppm) had only two correlations to the isoxazole (C4 and C5). Correlation between H10 and C3 was not observed because these atoms were four bonds apart.
|
|
For M3, only one of the two methyl signals was observed at 1.56 and 1.61 ppm in the 1.6 to 2.2 ppm region, and a new methylene signal appeared at 4.21 and 4.22 ppm, suggesting that one of the methyl group was hydroxylated. The HMBC spectrum of M3 (Fig. 5B) indicated that the methyl protons at 1.56 and 1.61 ppm had correlations to C3, C4, and C5 on the isoxazole ring, whereas the methylene protons at 4.21 and 4.22 ppm had only two correlations to the isoxazole (C4 and C5). These results suggested that position 9 was the methyl group at 1.56 and 1.61 ppm and that position 10 was the methylene group at 4.21 and 4.22 ppm. Therefore, M3 was identified as a hydroxy derivative of BMS-1, and all the other NMR data (Table 3) were consistent with the assigned structure. In the following discussion, M3 was referred to as 5-methylhydroxyisoxazole, and the reaction was termed isoxazole 5-methylhydroxylation.
In Vitro Metabolism of BMS-1 Analogs. The above data suggested that BMS-1 was subject to high metabolic clearance via CYP3A-mediated isoxazole 5-methylhydroxylation in monkey and human liver microsomes. To improve metabolic stability, the isoxazole ring was modified at both the 4- and 5-positions. Analogs with modifications at the 5-position were not evaluated for their metabolic and pharmacokinetic properties because of a marked loss of pharmacological activity. Instead, analogs with the halogenation at the 4-position were chosen, and as shown in Fig. 1, BMS-2 was a 4-chlorinated analog of BMS-1, whereas BMS-3 was a 4-fluorinated derivative.
Incubations of BMS-2 and BMS-3 in monkey and human liver microsomes revealed the following order of metabolic instability judged by percentage loss of parent compounds: BMS-1 > BMS-2 > BMS-3 (Table 1). As with the increase in stability, the formation of the 5-methylhydroxyisoxazole metabolites were also dramatically decreased for BMS-2 and BMS-3 relative to BMS-1, based on metabolite/parent ratios (Table 1). Both BMS-2 and BMS-3 were relatively stable in human liver microsomes as compared with those of monkey, and the formation of the 5-methylhydroxyisoxazole was further reduced after halogenation. In the presence of ketoconazole, metabolism of BMS-2 and BMS-3 in human liver microsomes was inhibited by 72 and 53%, respectively, based on percentage loss of parent (Table 2).
Incubations with cDNA-expressed human P450 enzymes indicated that BMS-2 and BMS-3 could be metabolized by both CYP3A4 and CYP2C9 (Table 2). Mass spectral analysis also suggested that CYP2C9 primarily catalyzed the hydroxylation and N-dealkylation on the 4'-moiety, and such reactions became more predominant for BMS-3. On the other hand, the formation of the 5-methylhyroxylisoxazole was catalyzed primarily by CYP3A4, and this reaction was reduced by halogenation at the 4-position.
In monkey and human liver microsomes, similar metabolite profiles were observed for BMS-2 and BMS-3 as previously shown for BMS-1. Mass spectral analyses showed that the isoxazole 5-hydroxylation was again the key metabolic pathway for these compounds in both monkey and human liver microsomes (Table 1), albeit at lower levels relative to BMS-1.
Evaluation of the Metabolic Activity in Monkeys Using Midazolam as a Probe Substrate. To obtain background information on potential impacts on pharmacokinetics, midazolam was administered p.o. to the three monkeys used throughout the present study. Plasma concentration versus time curves, as well as AUC and Cmax values, of midazolam are summarized in Fig. 6A and Table 4. Systemic exposure of midazolam varied significantly in these three monkeys (approximately 5-fold) after p.o. administration. Monkey-441 gave the highest systemic exposure to midazolam as reflected by AUC and Cmax values, whereas monkey-144 resulted in the lowest drug exposure and monkey-532 had intermediate exposure. When subject to in vitro incubations with monkey liver microsomes, midazolam (10 µM) showed a high turnover rate, with greater than 60% loss during the first 5 min of incubation. These data suggested that the interanimal variability in midazolam exposure was likely to be caused by the variability in CYP3A activities among these monkeys.
|
|
Pharmacokinetics of BMS-1, BMS-2, and BMS-3 in Monkeys. When BMS-1 was administered p.o. to the same three monkeys in a solution, significant variability in systemic exposure was observed (Fig. 6B; Table 4). Interestingly, the pattern of the interanimal variability was identical to that of midazolam in these monkeys, with the exposure following the order of monkey-441 > monkey-532 > monkey-144. In fact, plasma concentrations of BMS-1 were so low in monkey-144, the compound could only be detected in plasma at 15 min after dosing. Among these monkeys, the Cmax and AUC values of BMS-1 displayed marked variability (>19-fold).
When BMS-2 and BMS-3 were administered p.o. in solution to the same three monkeys, the same pattern of interanimal variability observed for midazolam and BMS-1 was again repeated for both BMS-2 and BMS-3 (Fig. 6, C and D), with systemic exposure following the order of monkey-441 > monkey-532 > monkey-144. For both BMS-2 and BMS-3, monkey-144 again gave lower values of Cmax and/or AUC, whereas the highest Cmax and AUC values were seen in monkey-441 (Table 4). For BMS-2, variability in Cmax and AUC remained high, even though metabolic stability was markedly improved as compared with BMS-1 (Table 1). For BMS-3, the degree of interanimal variability was dramatically reduced as compared with midazolam, BMS-1, and BMS-2, consistent with the reduced in vitro metabolism in monkey liver microsomes (Table 1).
Pharmacokinetic parameters of BMS-3 in monkeys after i.v. and p.o. administration are summarized in Table 5. BMS-3 had a moderate bioavailability of 19% in monkeys after p.o. administration. Time to reach maximum plasma concentration (Tmax) after p.o. administration was 2.0 h after the oral solution dose, suggesting rapid absorption. The compound exhibited an elimination half-life of 2.0 h after i.v. administration, with a plasma clearance of 27 ml/min/kg.
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
); coadministration of ketoconazole (a CYP3A4 inhibitor) can increase the AUC of midazolam given p.o. by up to 15-fold (Olkkola et al., 1994). Because CYP3A4-mediated drug-drug interactions frequently dictate a drug's clinical outcome and commercial value, it is desirable that CYP3A4 substrates, especially those metabolized predominantly by the enzyme, be identified and CYP3A4-mediated metabolism be minimized as early as possible during the drug discovery stage (Kumar and Surapaneni, 2001). In the present report, we describe a combination of oral pharmacokinetic evaluation in monkeys and in vitro drug metabolism studies that enabled the redesign of a lead compound and allowed identification of an analog that minimized both CYP3A-mediated metabolism and interanimal variability.
Orthologs of the human CYP3A4 enzyme are widely expressed in animal species. Although dissimilarities do exist between laboratory animals and humans in drug-metabolizing enzymes (Zuber et al., 2002), rodents (mouse and rat) and nonrodent species (monkey and dog) are often used to evaluate pharmacokinetics of new chemical entities. For cynomolgus monkeys, a number of genes encoding P450 enzymes have been cloned, and their DNA sequences are >90% similar to those in humans (Mankowski et al., 1999). Activities of drug-metabolizing enzymes in cynomolgus monkeys, particularly CYP3A, are generally higher than those found in humans (Sharer et al., 1995; Guengerich, 1997a; Shimada et al., 1997). As such, the cynomolgus monkey is thought to be a suitable yet stringent species to evaluate CYP3A-mediated metabolism (Chiou and Buehler, 2002). The utility of cynomolgus monkeys during the drug discovery stage has been explored for predicting oral absorption and drug-drug interactions (Ward et al., 2001, 2004; Kanazu et al., 2004).
Midazolam is a well established human CYP3A4 substrate and is often used to assess CYP3A activities in vivo (Thummel et al., 1994). Midazolam is generally not considered as a substrate of efflux pumps such as P-glycoprotein (Polli et al., 2001). Pharmacokinetic parameters of midazolam are principally governed by CYP3A-mediated metabolism (Thummel et al., 1996; Tolle-Sander et al., 2003). As reported by Kanazu et al. (2004), metabolism of midazolam in liver microsomes from cynomolgus monkeys is inhibited by anti-human CYP3A4 antiserum and chemical inhibitors; coadministration of ketoconazole resulted in marked increases in midazolam AUC values. These data suggest that metabolic clearance of midazolam in cynomolgus monkeys is also dependent on the CYP3A enzyme. Therefore, midazolam was used as a reference CYP3A substrate in the present study to obtain baseline information about CYP3A activity in the monkeys used. As shown in Table 4 and Fig. 6, the three monkeys showed variable oral pharmacokinetic profile of midazolam, indicating in vivo differences in the CYP3A activity.
During lead optimization for dual angiotensin and endothelin receptor antagonists, a liver microsomal assay was used extensively to generate not only quick and useful data on metabolic stability but also information such as enzyme inhibition and metabolite profiling. In this system, BMS-1 was found to be rapidly and extensively metabolized (metabolic clearance, monkey > human). Ketoconazole (2 µM) inhibited the metabolism of BMS-1, indicating that CYP3A was involved in the metabolism of BMS-1. The involvement of CYP3A was confirmed using cDNA-expressed human P450 enzymes, where only CYP3A4 showed marked turnover of BMS-1. The impact of CYP3A-mediated metabolism was further reflected in variable pharmacokinetics in monkeys. Administration p.o. of BMS-1 to monkeys resulted in marked interanimal variability in exposure, and the variability pattern was identical to that of midazolam, suggesting that CYP3A-mediated metabolism might be the contributing factor. Although the observed interanimal variability in monkeys could be caused by other factors, such as variable gastrointestinal emptying and transit time, as well as uptake and/or efflux transport processes, several lines of evidence suggest that CYP3A-mediated metabolism plays a major role. First, the variability pattern of BMS-1 exposure was identical to that of midazolam in these monkeys. Because midazolam is an appropriate in vivo probe for CYP3A activity (Thummel et al., 1994; Kanazu et al., 2004), the variability in midazolam exposure was therefore indicative of the variability in CYP3A activity (both intestinal and hepatic) in these monkeys. In addition, BMS-1 was found to be a CYP3A substrate for both monkey and human enzyme with a high metabolic clearance in monkeys. Furthermore, BMS-1 had adequate aqueous solubility (dosed as solution) and high permeability in the Caco-2 cells (Pc = 137 nm/s). Therefore, efflux transport of BMS-1 (if any) in the gastrointestinal tract of monkeys would be saturated at the dose used for oral pharmacokinetic evaluation. Even though the monkey is a more stringent preclinical species for pharmacokinetic evaluation because of higher levels of CYP3A activity (Sharer et al., 1995; Chiou and Buehler, 2002), a compound with low and variable exposure in monkeys as a result of extensive CYP3A-mediated metabolism would indicate a similar outcome in humans. In addition, a sole human CYP3A4 substrate would carry inevitable risks for significant drug-drug interactions when coadministered with potent CYP3A4 inhibitors (Gibbs and Hosea, 2003). For this reason, it is desirable to minimize CYP3A-mediated metabolism of BMS-1, with the expectation that improvement in BMS-1's metabolic stability and minimization of CYP3A-mediated metabolism would reduce interanimal variability in monkeys.
With this in mind, additional in vitro metabolism studies were carried out to identify metabolic soft spots for BMS-1. Mass spectral analysis, in conjunction with NMR, indicated that M3 was a hydroxylated metabolite of BMS-1 that was formed via the hydroxylation of the 5-methyl group on the isoxazole ring. Formation of M3 in cDNA-expressed human CYP3A4 suggested that the 5-methyl hydroxylation was a site-specific CYP3A-mediated metabolism, which was further supported by the inhibition of microsomal metabolism by ketoconazole. Therefore, structural elucidation of M3 allowed medicinal chemistry to explore structure-activity relationship to reduce the propensity of site-specific CYP3A-mediated metabolism seen in BMS-1. As represented by BMS-2 and BMS-3, halogenation of the isoxazole ring not only improved in vitro metabolic stability in liver microsomes but also minimized site-specific CYP3A-mediated metabolism. This is possibly because of electron withdrawing effects of the ring halogen, making H-atom abstraction from the ring-methyl carbon less favorable. In addition, halogenation could also produce a change in the steric interaction of the molecule with the P450 active site. Interestingly, these structural modifications were accompanied by the involvement of other metabolic pathways such as CYP2C9 metabolism (Table 2). When subject to pharmacokinetic evaluation in the same three monkeys, BMS-2 still showed considerable interanimal variability in drug exposure, suggesting that chlorination was not sufficient to overcome significant in vivo differences in CYP3A-mediated metabolism among these three monkeys. In contrast, BMS-3 (a fluorinated analog of BMS-1) had a much lower degree of interanimal variability in systemic exposure. These results confirmed the hypothesis that improving overall metabolic stability and reducing the fractional clearance through the CYP3A pathway would at least in part reduce the interanimal variability.
Although the variability of BMS-3 exposure was greatly reduced when compared with BMS-1 and BMS-2, the compound still exhibited some interanimal variability with the same patterns as exhibited by other substrates in these monkeys. One possible explanation is that even though in vitro metabolism of BMS-3 was markedly reduced, CYP3A was still a major metabolic enzyme, and the CYP3A-mediated metabolism was probably higher in vivo, resulting in differences in metabolic clearance at both the intestinal and hepatic level in these monkeys. In addition, BMS-3 had an average plasma clearance of 27 ml/min/kg after i.v. administration, suggesting that BMS-3 was subject to high in vivo clearance (Davies and Morris, 1993). It could be postulated that in vivo clearance of BMS-3 might also be governed by mechanism(s) other than CYP3A-mediated metabolism. The clearance pathways of BMS-3 in vivo will be the subject of a separate investigation.
In summary, the present report summarizes an integrated strategy for understanding the impact of CYP3A-mediated metabolism on the pharmacokinetic parameters of test compounds in cynomolgus monkeys. This strategy led to the identification of analogs with greatly improved pharmacokinetic profiles in monkeys, which should predict a similar outcome in humans. In particular, liver microsomes and cDNA-expressed human P450 enzymes were used to model CYP3A-mediated in vitro metabolism, in conjunction with in vivo studies in monkeys for CYP3A activity with oral midazolam and pharmacokinetic evaluation. Structural elucidation of the major metabolite of BMS-1 served as the key turning point during lead optimization. As a result, a qualitative in vitro to in vivo correlation was observed, and subsequent structural modification targeting CYP3A-mediated metabolism led to the reduction of interanimal variability in monkeys. Our study showed that monkey can be a very useful preclinical model for pharmacokinetic evaluation, although it may represent a more stringent measure with respect to P450 activity in general and CYP3A activity in particular. For this reason, it is important to fully understand the in vitro and in vivo metabolic characteristics of new compounds to allow the best prediction of absorption, distribution, metabolism, and elimination characteristics in humans. Finally, caution should be exercised for a sole CYP3A substrate as a drug candidate, and in-depth evaluation must be made with respect to the therapeutic indication, safety margins, and potential drug-drug interactions.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
ABBREVIATIONS: P450, cytochrome P450; BMS-1, 2-{butyryl-[2'-(4,5-dimethyl-isoxazol-3-ylsulfamoyl)-biphenyl-4-ylmethyl]-amino}-N-isopropyl-3-methyl-butyramide; BMS, Bristol-Myers Squibb; HPLC, high-performance liquid chromatography; LC/MS, liquid chromatography/mass spectrometry; LC/MS/MS, liquid chromatography/tandem mass spectrometry; DMSO, dimethyl sulfoxide; HMBC, heteronuclear multiple bond coherence; IS, internal standard; BMS-2, 2-{butyryl-[2'-(4-chloro-5-methyl-isoxazol-3-ylsulfamoyl)-biphenyl-4-ylmethyl]-amio}-N-isopropyl-3-methyl-butyramide; BMS-3, 2-{butyryl-[2'-(4-fluoro-5-methyl-isoxazol-3-ylsulfamoyl)-biphenyl-4-ylmethyl]-amino}-N-isopropyl-3-methylbutyramide; AUC, area under the plasma concentration-time curve.
Address correspondence to: Hongjian Zhang, Metabolism and Pharmacokinetics, Pharmaceutical Candidate Optimization, Bristol-Myers Squibb Company, P.O. Box 4000, Princeton, NJ 08543. E-mail: hongjian.zhang{at}bms.com
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
 |
|
 |
 |
 |
